X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Freverse_proxy.ex;h=91cf1bba33006f9947421e5799ad869144cf68de;hb=9c7178286116d61a565fcba61ec64b20fec3a28a;hp=285d57309ea76a488ba14a51508d14daf76c3f10;hpb=f4e2595592ccca6cedd64669baef7bdd2a6547d0;p=akkoma

diff --git a/lib/pleroma/reverse_proxy.ex b/lib/pleroma/reverse_proxy.ex
index 285d57309..91cf1bba3 100644
--- a/lib/pleroma/reverse_proxy.ex
+++ b/lib/pleroma/reverse_proxy.ex
@@ -1,22 +1,27 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.ReverseProxy do
-  alias Pleroma.HTTP
-
-  @keep_req_headers ~w(accept user-agent accept-encoding cache-control if-modified-since) ++
-                      ~w(if-unmodified-since if-none-match if-range range)
-  @resp_cache_headers ~w(etag date last-modified cache-control)
+  @range_headers ~w(range if-range)
+  @keep_req_headers ~w(accept accept-encoding cache-control if-modified-since) ++
+                      ~w(if-unmodified-since if-none-match) ++ @range_headers
+  @resp_cache_headers ~w(etag date last-modified)
   @keep_resp_headers @resp_cache_headers ++
-                       ~w(content-type content-disposition content-encoding content-range) ++
-                       ~w(accept-ranges vary)
+                       ~w(content-length content-type content-disposition content-encoding) ++
+                       ~w(content-range accept-ranges vary expires)
   @default_cache_control_header "public, max-age=1209600"
   @valid_resp_codes [200, 206, 304]
   @max_read_duration :timer.seconds(30)
   @max_body_length :infinity
+  @failed_request_ttl :timer.seconds(60)
   @methods ~w(GET HEAD)
 
+  @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
+
+  def max_read_duration_default, do: @max_read_duration
+  def default_cache_control_header, do: @default_cache_control_header
+
   @moduledoc """
   A reverse proxy.
 
@@ -33,9 +38,6 @@ defmodule Pleroma.ReverseProxy do
   * request: `#{inspect(@keep_req_headers)}`
   * response: `#{inspect(@keep_resp_headers)}`
 
-  If no caching headers (`#{inspect(@resp_cache_headers)}`) are returned by upstream, `cache-control` will be
-  set to `#{inspect(@default_cache_control_header)}`.
-
   Options:
 
   * `redirect_on_failure` (default `false`). Redirects the client to the real remote URL if there's any HTTP
@@ -48,21 +50,16 @@ defmodule Pleroma.ReverseProxy do
   * `max_read_duration` (default `#{inspect(@max_read_duration)}` ms): the total time the connection is allowed to
   read from the remote upstream.
 
+  * `failed_request_ttl` (default `#{inspect(@failed_request_ttl)}` ms): the time the failed request is cached and cannot be retried.
+
   * `inline_content_types`:
     * `true` will not alter `content-disposition` (up to the upstream),
     * `false` will add `content-disposition: attachment` to any request,
     * a list of whitelisted content types
 
-    * `keep_user_agent` will forward the client's user-agent to the upstream. This may be useful if the upstream is
-    doing content transformation (encoding, …) depending on the request.
-
   * `req_headers`, `resp_headers` additional headers.
 
-  * `http`: options for [hackney](https://github.com/benoitc/hackney).
-
   """
-  @default_hackney_options []
-
   @inline_content_types [
     "image/gif",
     "image/jpeg",
@@ -80,9 +77,9 @@ defmodule Pleroma.ReverseProxy do
   import Plug.Conn
 
   @type option() ::
-          {:keep_user_agent, boolean}
-          | {:max_read_duration, :timer.time() | :infinity}
+          {:max_read_duration, :timer.time() | :infinity}
           | {:max_body_length, non_neg_integer() | :infinity}
+          | {:failed_request_ttl, :timer.time() | :infinity}
           | {:http, []}
           | {:req_headers, [{String.t(), String.t()}]}
           | {:resp_headers, [{String.t(), String.t()}]}
@@ -93,10 +90,7 @@ defmodule Pleroma.ReverseProxy do
   def call(_conn, _url, _opts \\ [])
 
   def call(conn = %{method: method}, url, opts) when method in @methods do
-    hackney_opts =
-      @default_hackney_options
-      |> Keyword.merge(Keyword.get(opts, :http, []))
-      |> HTTP.process_request_options()
+    client_opts = Keyword.get(opts, :http, [])
 
     req_headers = build_req_headers(conn.req_headers, opts)
 
@@ -107,31 +101,52 @@ defmodule Pleroma.ReverseProxy do
         opts
       end
 
-    with {:ok, code, headers, client} <- request(method, url, req_headers, hackney_opts),
-         :ok <- header_length_constraint(headers, Keyword.get(opts, :max_body_length)) do
-      response(conn, client, url, code, headers, opts)
+    with {:ok, nil} <- @cachex.get(:failed_proxy_url_cache, url),
+         {:ok, status, headers, body} <- request(method, url, req_headers, client_opts),
+         :ok <-
+           header_length_constraint(
+             headers,
+             Keyword.get(opts, :max_body_length, @max_body_length)
+           ) do
+      conn
+      |> put_private(:proxied_url, url)
+      |> response(body, status, headers, opts)
     else
-      {:ok, code, headers} ->
-        head_response(conn, url, code, headers, opts)
+      {:ok, true} ->
+        conn
+        |> put_private(:proxied_url, url)
+        |> error_or_redirect(500, "Request failed", opts)
         |> halt()
 
-      {:error, {:invalid_http_response, code}} ->
-        Logger.error("#{__MODULE__}: request to #{inspect(url)} failed with HTTP status #{code}")
+      {:ok, status, headers} ->
+        conn
+        |> put_private(:proxied_url, url)
+        |> head_response(status, headers, opts)
+        |> halt()
+
+      {:error, {:invalid_http_response, status}} ->
+        Logger.error(
+          "#{__MODULE__}: request to #{inspect(url)} failed with HTTP status #{status}"
+        )
+
+        track_failed_url(url, status, opts)
 
         conn
+        |> put_private(:proxied_url, url)
         |> error_or_redirect(
-          url,
-          code,
-          "Request failed: " <> Plug.Conn.Status.reason_phrase(code),
+          status,
+          "Request failed: " <> Plug.Conn.Status.reason_phrase(status),
           opts
         )
         |> halt()
 
       {:error, error} ->
         Logger.error("#{__MODULE__}: request to #{inspect(url)} failed: #{inspect(error)}")
+        track_failed_url(url, error, opts)
 
         conn
-        |> error_or_redirect(url, 500, "Request failed", opts)
+        |> put_private(:proxied_url, url)
+        |> error_or_redirect(500, "Request failed", opts)
         |> halt()
     end
   end
@@ -142,86 +157,52 @@ defmodule Pleroma.ReverseProxy do
     |> halt()
   end
 
-  defp request(method, url, headers, hackney_opts) do
+  defp request(method, url, headers, opts) do
     Logger.debug("#{__MODULE__} #{method} #{url} #{inspect(headers)}")
     method = method |> String.downcase() |> String.to_existing_atom()
 
-    case :hackney.request(method, url, headers, "", hackney_opts) do
-      {:ok, code, headers, client} when code in @valid_resp_codes ->
-        {:ok, code, downcase_headers(headers), client}
+    opts = opts ++ [receive_timeout: @max_read_duration]
+
+    case Pleroma.HTTP.request(method, url, "", headers, opts) do
+      {:ok, %Tesla.Env{status: status, headers: headers, body: body}}
+      when status in @valid_resp_codes ->
+        {:ok, status, downcase_headers(headers), body}
 
-      {:ok, code, headers} when code in @valid_resp_codes ->
-        {:ok, code, downcase_headers(headers)}
+      {:ok, %Tesla.Env{status: status, headers: headers}} when status in @valid_resp_codes ->
+        {:ok, status, downcase_headers(headers)}
 
-      {:ok, code, _, _} ->
-        {:error, {:invalid_http_response, code}}
+      {:ok, %Tesla.Env{status: status}} ->
+        {:error, {:invalid_http_response, status}}
 
       {:error, error} ->
         {:error, error}
     end
   end
 
-  defp response(conn, client, url, status, headers, opts) do
-    result =
-      conn
-      |> put_resp_headers(build_resp_headers(headers, opts))
-      |> send_chunked(status)
-      |> chunk_reply(client, opts)
-
-    case result do
-      {:ok, conn} ->
-        halt(conn)
-
-      {:error, :closed, conn} ->
-        :hackney.close(client)
-        halt(conn)
-
-      {:error, error, conn} ->
-        Logger.warn(
-          "#{__MODULE__} request to #{url} failed while reading/chunking: #{inspect(error)}"
-        )
+  defp response(conn, body, status, headers, opts) do
+    Logger.debug("#{__MODULE__} #{status} #{conn.private[:proxied_url]} #{inspect(headers)}")
 
-        :hackney.close(client)
-        halt(conn)
-    end
-  end
-
-  defp chunk_reply(conn, client, opts) do
-    chunk_reply(conn, client, opts, 0, 0)
+    conn
+    |> put_resp_headers(build_resp_headers(headers, opts))
+    |> send_resp(status, body)
   end
 
-  defp chunk_reply(conn, client, opts, sent_so_far, duration) do
-    with {:ok, duration} <-
-           check_read_duration(
-             duration,
-             Keyword.get(opts, :max_read_duration, @max_read_duration)
-           ),
-         {:ok, data} <- :hackney.stream_body(client),
-         {:ok, duration} <- increase_read_duration(duration),
-         sent_so_far = sent_so_far + byte_size(data),
-         :ok <- body_size_constraint(sent_so_far, Keyword.get(opts, :max_body_size)),
-         {:ok, conn} <- chunk(conn, data) do
-      chunk_reply(conn, client, opts, sent_so_far, duration)
-    else
-      :done -> {:ok, conn}
-      {:error, error} -> {:error, error, conn}
-    end
-  end
+  defp head_response(conn, status, headers, opts) do
+    Logger.debug("#{__MODULE__} #{status} #{conn.private[:proxied_url]} #{inspect(headers)}")
 
-  defp head_response(conn, _url, code, headers, opts) do
     conn
     |> put_resp_headers(build_resp_headers(headers, opts))
-    |> send_resp(code, "")
+    |> send_resp(status, "")
   end
 
-  defp error_or_redirect(conn, url, code, body, opts) do
+  defp error_or_redirect(conn, status, body, opts) do
     if Keyword.get(opts, :redirect_on_failure, false) do
       conn
-      |> Phoenix.Controller.redirect(external: url)
+      |> Phoenix.Controller.redirect(external: conn.private[:proxied_url])
       |> halt()
     else
       conn
-      |> send_resp(code, body)
+      |> send_resp(status, body)
       |> halt
     end
   end
@@ -250,20 +231,19 @@ defmodule Pleroma.ReverseProxy do
     headers
     |> downcase_headers()
     |> Enum.filter(fn {k, _} -> k in @keep_req_headers end)
-    |> (fn headers ->
-          headers = headers ++ Keyword.get(opts, :req_headers, [])
-
-          if Keyword.get(opts, :keep_user_agent, false) do
-            List.keystore(
-              headers,
-              "user-agent",
-              0,
-              {"user-agent", Pleroma.Application.user_agent()}
-            )
-          else
-            headers
-          end
-        end).()
+    |> build_req_range_or_encoding_header(opts)
+    |> Keyword.merge(Keyword.get(opts, :req_headers, []))
+  end
+
+  # Disable content-encoding if any @range_headers are requested (see #1823).
+  defp build_req_range_or_encoding_header(headers, _opts) do
+    range? = Enum.any?(headers, fn {header, _} -> Enum.member?(@range_headers, header) end)
+
+    if range? && List.keymember?(headers, "accept-encoding", 0) do
+      List.keydelete(headers, "accept-encoding", 0)
+    else
+      headers
+    end
   end
 
   defp build_resp_headers(headers, opts) do
@@ -271,21 +251,22 @@ defmodule Pleroma.ReverseProxy do
     |> Enum.filter(fn {k, _} -> k in @keep_resp_headers end)
     |> build_resp_cache_headers(opts)
     |> build_resp_content_disposition_header(opts)
-    |> (fn headers -> headers ++ Keyword.get(opts, :resp_headers, []) end).()
+    |> Keyword.merge(Keyword.get(opts, :resp_headers, []))
   end
 
   defp build_resp_cache_headers(headers, _opts) do
     has_cache? = Enum.any?(headers, fn {k, _} -> k in @resp_cache_headers end)
-    has_cache_control? = List.keymember?(headers, "cache-control", 0)
 
     cond do
-      has_cache? && has_cache_control? ->
-        headers
-
       has_cache? ->
-        # There's caching header present but no cache-control -- we need to explicitely override it
-        # to public as Plug defaults to "max-age=0, private, must-revalidate"
-        List.keystore(headers, "cache-control", 0, {"cache-control", "public"})
+        # There's caching header present but no cache-control -- we need to set our own
+        # as Plug defaults to "max-age=0, private, must-revalidate"
+        List.keystore(
+          headers,
+          "cache-control",
+          0,
+          {"cache-control", @default_cache_control_header}
+        )
 
       true ->
         List.keystore(
@@ -351,30 +332,14 @@ defmodule Pleroma.ReverseProxy do
 
   defp header_length_constraint(_, _), do: :ok
 
-  defp body_size_constraint(size, limit) when is_integer(limit) and limit > 0 and size >= limit do
-    {:error, :body_too_large}
-  end
-
-  defp body_size_constraint(_, _), do: :ok
-
-  defp check_read_duration(duration, max)
-       when is_integer(duration) and is_integer(max) and max > 0 do
-    if duration > max do
-      {:error, :read_duration_exceeded}
-    else
-      {:ok, {duration, :erlang.system_time(:millisecond)}}
-    end
-  end
-
-  defp check_read_duration(_, _), do: {:ok, :no_duration_limit, :no_duration_limit}
-
-  defp increase_read_duration({previous_duration, started})
-       when is_integer(previous_duration) and is_integer(started) do
-    duration = :erlang.system_time(:millisecond) - started
-    {:ok, previous_duration + duration}
-  end
+  defp track_failed_url(url, error, opts) do
+    ttl =
+      unless error in [:body_too_large, 400, 204] do
+        Keyword.get(opts, :failed_request_ttl, @failed_request_ttl)
+      else
+        nil
+      end
 
-  defp increase_read_duration(_) do
-    {:ok, :no_duration_limit, :no_duration_limit}
+    @cachex.put(:failed_proxy_url_cache, url, true, ttl: ttl)
   end
 end